Humanization of mAbs supported by molecular simulation

Human therapies with non-human derived antibodies (Ab) lead to adverse side effects during repeated application in patients. Especially reduced serum half live and activation of immunological reactions diminish the therapeutic efficacy and are responsible for rejection of the drug by induction of non specific anti-drug responses. To overcome these problems, humanization approaches for murine monoclonal antibodies (mAb) have been developed using different techniques to reduce the immunogenicity of mouse Abs. Because numerous powerful mouse-mAbs are developed in scientific projects, we face an increased demand to improve humanization approaches. There are many mAbs that are tested positively in preclinical trials and therefore represent valuable candidates for human application. A state of the art method for the humanization process is to graft the murine complementarity-determining regions (CDRs) of the murine Ab onto human frameworks of variable regions (FR) and analyze the newly generated Ab based on structural models derived from existing Abs. After this sequential analysis of single amino acids, individual point mutations are considered in the FR of the newly generated Abs. Finally a couple of different Ab variants are expressed and characterized by in vitro and in vivo methods. This loop of expression and testing is rather time consuming and often expensive. In this project we will go one step further and include dynamic molecular models to support predictions on the similarity between the original Ab and the newly generated humanized version. Homology models will be generated and subsequently subjected to molecular dynamics simulations. By using advanced and enhanced sampling techniques, we will select those humanized mutants with similar atomic structure and dynamics as the original murine Ab which will most likely still bind favorably to the target protein. In this context we aim to replace the common trial and error approach with a rational design to maintain a given structure. In this sense, new science will result from merging the two leading disciplines of the 21st century, namely biotechnology and bioinformatics, in an iterative, multidisciplinary setup. The computational approach will first be applied to a successfully humanized Ab to establish a proof-of-principle and subsequently be validated on the mouse derived Ab2/3H6, which has been extensively studied in the applicants lab. The project will also give an input to other scientific branches. Protein chemistry will profit from more accurate prediction of macromolecular structures, which will also help in the description of interactions with other substances like small molecule substrates, proteins or membranes. Moreover, our tools aim to generate artificial and novel proteins with improved therapeutic potential, reduced immunogenic side effects and increased half live for human application. Additionally, cell biology and physiology will benefit from recombinant cell culture clones secreting similar antibodies with defined protein structures.

Proteins applied in human therapy have to meet severe regulatory issues to meet the safety standards of the patients. This concerns the purity of the product and the absence of toxic byproducts which are mainly driven by the production process. Additional side effects can be determined by the protein when it differs from the naturally occurring sequence in the human body. Inflammatory reactions caused by the immune system or reduced half-life of the biopharmaceutical are typical side reactions. Monoclonal antibodies are indicated by a huge number of individual proteins which are often developed in lab animals and afterwards are modified to generate humanized monoclonal antibodies. This process is accomplished individually and supported by empirical knowledge gained from the literature to preserve the valuable efficacy of the molecule but also to diminish the discussed potential side reactions. In this project we have chosen a precise example (Ab2/3H6 is a mouse derived monoclonal antibody) to evaluate different humanization strategies and to merge the empirical knowledge from literature with a molecular simulation approach in close cooperation between the Department of Biotechnology and the Institute of Molecular Simulation. This will lead in the future to structural prediction of the monoclonal antibody with the advantage of faster development time and less effort in the wet lab. The humanization of antibodies remains a challenging task in the context of rational drug design. Superhumanization describes the direct transfer of the complementarity determining regions to a human germline framework, but this humanization approach often results in loss of binding affinity. In this project, we developed a new approach for predicting promising backmutation sites using molecular dynamics (MD) simulations of the model antibody Ab2/3H6. Differences in the framework regions lead to conformational changes in the complementarity determining regions, and thereby possibly to a loss of affinity. The simulation approach developed in the current project relies on extensive MD simulations of putative variants and an analysis of the similarity of the conformations of the binding interface between the wildtype, murine, antibody and the humanized variant. The simulation method was developed in close conjunction with novel specificity experiments. Binding properties of mAb variants were evaluated directly from crude supernatants and confirmed using established binding affinity assays for purified antibodies. Our approach provides access to the dynamical features of the actual binding sites of an antibody, based solely on the antibody sequence. Thus we do not need structural data on the antibody-antigen complex and circumvent cumbersome methods to assess binding affinities.